In a conventional PSTN (Public Switched Telephone Network, public switched telephone network) network, a 64 K bandwidth is generally provided, where a 3.4 K spectrum is used for transmitting a voice signal. Because a spectrum of people speaking may generally reach 7 K, distortion usually occurs in voice with a 3.4 K spectrum provided in the conventional PSTN network, which is also the reason why voice of people on the telephone is different from that in a practical environment. Compared with the conventional PSTN network, because an audio signal with a frequency reaching 7 K may be processed in a G.722 encoding/decoding manner, in order to solve a voice distortion problem, in an IP (Internet Protocol, Internet protocol) network, a voice solution based on the G.722 encoding/decoding manner is provided by many chip manufacturers.
As shown in FIG. 1, two parts of hardware are required to implement voice encoding/decoding that is based on G.722 in the prior art. One part is a POTS (Plain Old Telephone Service, plain old telephone service) subscriber board, and the subscriber board includes a Codec (codec)/SLIC (Subscriber Line Interface Circuit, subscriber line interface circuit); and the other part is a DSP (Digital Signal Processing, digital signal processing) chip. In a specific process of voice encoding, the DSP chip multiplies frequencies of two 8 K PCM (Pulse Code Modulation, pulse code modulation) signals into 16 K, and 16-K sampling is implemented through two time slots. In addition, a 16-K based processing mode is also used inside the DSP chip. The PCM signal with two time slots is restored to 16-K data, and then EC (echo cancel, echo cancel)/Tone Detect (tone detect) and encoding are performed on the 16-K data, and finally, an encoded signal is output in an RTP (Real-time Transport Protocol, real-time transport protocol) format. A voice decoding process is a reverse process of the voice encoding.
Since voice with a 7 K spectrum is not widely applied at present, and a current network mainly adopts the voice with the 3.4 K spectrum, the DSP chip applied in the current network generally does not support 16-K frequency multiplication and 16-K code stream processing, that is, products widely used in the current network cannot provide a function of encoding/decoding voice with a 7 K spectrum. The 16-K frequency multiplication needs to be supported by hardware inside the DSP chip. If a voice encoding/decoding implementation solution provided in the prior art is expected to be supported, the hardware inside the DSP chip in the current network needs to be replaced.